MR APs/mGig/CAT6a

[IainW]

Hello everyone,

We've performed several installations recently with Meraki MR44 access points connecting via CAT6a cable back to MS350/355 series switches.

In most cases, the AP and switch port negotiate a 2500Mbps connection as expected but we always find the odd AP that will auto-negotiate to 100Mbps.  The infrastructure cabling has been Fluke tested end-to-end and passed without a problem and if the AP is directly connected to the switch, 2500Mbps isn't an issue.

The problem seems to be extreme sensitivity around the Ethernet port on the AP and where the CAT6a cable is plugged in.  We are using angled plugs to mitigate any bends in the cable but it seems the slightest knock when fitting the AP to the ceiling brings the AP down. 

I just wondered if anyone else had experienced this and had any creative suggests for mitigation?

Cheers!

[DarrenOC]

Hi @IainW , I presume this is part of the DfE connect the classroom initiative?

how far are your cable runs out of interest?

[IainW]

Some are part of CtC, others are just normally funded commercial installations.

We've seen the same issue on runs as low as 15m through to 80+m.

All runs are below the 100m limit.

[DarrenOC]

How many APs are you loading per switch?  

[IainW]

Anywhere up to 8 on an MS350-24X.  Where I've seen the issue today, no more than 16 on an MS355-24X2.

[DarrenOC]

So you’re well within the device limitations on number of mgig ports.

nothing in the event log for reason applying low power?

[DarrenOC]

You're 100% confident in your cablers?  If the AP works directly connected to the switch then that rules out the AP and switch....only other thing it can be is the cabling....

[DarrenOC]

and to date I’ve not had a duff AP shipped to me.

[mikecarter]

I would also check any patch cables in use as well as the path of cable runs. Ensure none are too close to any fluorescent light ballasts.

I had a similar issue with all of the cable runs testing perfect, but a couple of faulty patch cables. 

[TBHPTL]

Define fluke tested.. were the cables runs simply pin mapped or was an actual certification done on the plant?  Permanent link or channel certified?

What OEM system of installation  did you  use and are all pieces of permanent or channel link compliant?   If perm link certified are your patch cables compliant width the OEM system of installation?

if something is negotiating at 10OMbps you are likely "missing" two pairs or there is noise on those two pairs.

[IainW]

Hello everyone.

Thanks for the comments so far.  To answer some of the questions and provide an update, the cabling in question is using U/FTP and has been channel tested/certified to TIA-CAT6a.  It's also worth noting that I'm not seeing this issue with all APs in this particular installation, it was around 10% of a 60+ MR44 deployment.

I'm confident the issue is where the cable meets the access point ethernet socket rather than loss on the cable itself.  The AP I was playing with yesterday appeared to be super sensitive - as in if the cable doesn't meet the AP ethernet socket at exactly the correct angle, it's 100Mbps only.  With the AP off the ceiling and ensuring there was no pressure on the ethernet jack resulted in a 2.5Gbps connection every time. Other MR44's in the same delivery appear to be much less sensitive. 

Ultimately, I think my problem is being caused by the sheer thickness of the CAT6a cable and the lack of space in the back of the AP to effectively allow the plug to sit in the socket without there being a slight angle on it and this is causing the pins to not make full contact.

[Speedbird1]

Seen this before and its always been cabling not being terminated correctly.

We used to terminate cat6a with Rj45 plugs and as you suggested plugged into directly into an AP. There just isn't enough room.

Now we always ask for all terminations to AP's be in patch boxes where a patch cable can be run to an AP. 

We have had 2 sites re-wired because the contractor said its better to terminate with RJ45 plugs. I think not.

[TBHPTL]

I suspect that your install team is manhandling the cables, not realizing the consequences. This should not happen if the install team is certified in the OEM's system of cable installation.

If you certified the channel that means its inclusive of patch cords on each end. if you take your cert machine and place it on the one of the offending channels what is the result?

Here is an example of a well intentioned  rookie Installer who thought this was permissible when installing the AP end of the cable into the cable well. This caused all sorts of issues from  100Mb connection to AP that would receive PoE but showed no signs of life. This is a Panduit CAT6A cable. Out of 40 APs installed the 12 this dude touched was hosed.

Installing cables with a boot requires that you take your time and be carful not to tweak the cable or else the internal spline will help your conductors  separate from the pins. I would recommend new patch cords WITHOUT boots and of the small diameter (28 gauge) variety. When installing remove the AP and carefully thread the patch through the mounting bracket taking care not to tweak the end. I would hold the body of the cable about 6 inches back and guide it into the well. It will in fact, be close quarters when installing the cable into the well. Your installer will have to be careful to not "abuse" the cable when pulling slack back to the mounting bracket

https://www.panduit.com/en/products/copper-systems/patch-cords-accessories/patch-cords/utp28x7bu.htm...

I don't have a cat6A at my desk but here is a sample of the Panduit  CAT6 standard vs a CAT 6 small diameter. The size difference is easily seen and makes installs a little more forgiving in tight quarters.